NEW YORK – Investigators from TGen, working with City of Hope, Mayo Clinic, and other clinical collaborators, are gearing up to clinically validate and prove utility for their liquid biopsy method to monitor patients with early-stage cancer as they are treated, allowing doctors to better predict if they need additional interventions or may be effectively cured.
The team described the technique — which researchers call TARDIS (targeted digital sequencing) — in Science Translational Medicine this week, reporting that it achieved what they believe to be a 100-fold improvement in sensitivity over the limits of other ctDNA detection approaches.
Investigators said they were able to reach 91 percent sensitivity in detecting tumor DNA mutations at mutant allele fractions of 3 in 10,000, and 53 percent sensitivity down to a MAF of 3 in 100,000, both with 96 percent specificity.
Unlike the liquid biopsy technologies that are already being applied to guide precision oncology in advanced cancer patients, the developers of TARDIS are aiming to improve and personalize the treatment of patients with early-stage, non-metastatic disease.
In this setting, depending on existing risk factors, individuals might be treated with surgery alone, with neoadjuvant drug treatment and then surgery, and then with various potential adjuvant strategies.
The difficulty is that some of these patients go on to see their disease recur while others do not, and current tools do a poor job at helping to identify which group is which. This means that adjuvant treatments may be applied to patients that don’t need them or withheld from those who would benefit.
In the drug development arena, this also makes it difficult to create and test new adjuvant therapies, because potential test cohorts are muddied by large numbers of individuals who don't stand to benefit from an experimental therapy, making it hard to prove efficacy of a drug or other intervention.
The TGen team is not alone in hoping to apply liquid biopsy approaches to address this problem. Some companies that started their liquid biopsy activities in the metastatic cancer space are now eyeing patient monitoring. Other firms, like Natera, have launched from the get-go with monitoring-directed assays.
Muhammed Murtaza, the new TARDIS study's senior author and the co-director of TGen's Center for Noninvasive Diagnostics, said in an interview that as techniques of this kind are just beginning to enter the commercial market now, it will take time for the clinical community, regulators, and payors to figure out just how sensitive such tools need to be.
But according to Murtaza, there is evidence that while described methods from other groups or companies can successfully track changes in circulating biomarkers during or after treatment, they tend to hit a wall at which ctDNA becomes undetectable to them, even in patients who continue to have observable disease on imaging.
"When we started wanting to do this a few years ago, one of the first thing we realized was that ctDNA levels were much, much lower in patients who have non-metastatic disease than in those with metastatic disease," Murtaza said. "So in that way the challenge here really extends beyond the performance of any single method."
"If you are looking at tumor fraction in the neighborhood of 0.1 percent [in patients] at presentation, and you take 10 ml blood, you have roughly 6,000 copies of the genome total. Any single mutation you want to look for, you can at best — not accounting for any inefficiencies — interrogate about six molecules from the tumor. And everything else you do to treat the patient decreases [that concentration] from there," he explained.
A breakthrough in this has been techniques like the one Natera now sells and the TGen team has published, which use tumor tissue sequencing to identify and curate a set of patient-specific mutations, which can then be tracked in blood in parallel. Having a handful of different targets increases the chance of picking up any single one as allele fractions for each alteration drop lower and lower.
In broad strokes, TARDIS involves this basic concept of multiplexed analysis of patient-specific cancer mutations in circulating cell-free DNA. For each individual, a bespoke, or patient-specific panel of mutations is created, using data from tumor tissue exome sequencing.
Investigators have been experimenting with this type of approach for several years already. But according to Murtaza and colleagues, while other techniques that create bespoke assays have significantly improved sensitivity, they still have limitations. When patients are treated with neoadjuvant drugs, especially, he and his coauthors wrote, ctDNA levels can readily drop below the limit of detection of various available methods.
"PCR-based approaches have been limited by the high background error rates observed," the study authors wrote. And while barcoding strategies can help overcome the issue of background errors, it also limits molecular conversion and multiplexing capacity.
In contrast, the researchers claim, ligation-based sequencing library preparation enables a wider analysis of the genome but has limited molecular conversion and loses up to 90 percent of template DNA molecules due to inefficient ligation. Incorporating many more mutations helps, but it limits applicability to tumor types where higher mutation rates are prevalent.
"An even wider analysis can be performed by WGS of plasma DNA … [but] WGS of plasma DNA at the required depth of coverage remains prohibitively expensive for a ctDNA test that may be repeated multiple times during clinical follow-up," the authors wrote.
Murtaza said that on a technical level, TARDIS combines some strengths of both amplicon and ligation-based methods. The combination helps to minimize loss of template DNA molecules, while also reducing background error issues, and allowing multiplexing into the hundreds, as demonstrated in the study this week.
Additionally, Murtaza said that the approach involves an optimized protocol "to the point where we can switch content for every patient at the scale of dozens — potentially hundreds of mutations." In the study, the authors wrote that using automated informatics pipelines, they can design and synthesize a TARDIS assay for each patient within one to two weeks after surgical removal and exome sequencing of a patient's tumor (something that is performed routinely for patients seen at the institutions that participated in the study.
Testing the method for their report this week, the investigators first created a multiplexed panel targeting eight mutations present in a commercially available reference sample from SeraCare.
Analyzing a total of 93 replicates from 1 percent to 0.031 percent allele fractions, alongside 16 wild-type samples, the team reported a sample-level sensitivity of 100 percent down to 0.125 percent AF, 87.5 percent for 0.063 percent AF samples, and 78.6 percent at 0.031 percent AF.
They followed this up with another experiment using 16 mutations in 56 reference sample replicates and were able to detect tumor DNA in all 1 percent AF samples, 87.5 percent of 0.03 percent AF samples, and 53 percent at 0.003 percent AF.
After various other analytical experiments, the group finally applied their approach to a set of patient samples, from 33 individuals with stage I to stage III breast cancer, of whom 22 were treated with neoadjuvant therapy.
Among the cohort, 24 had stage II disease and 30 of the 33 had invasive ductal carcinoma. The majority were diagnosed with ER+ HER2− tumors, with a few HER2+ and TNBC cases. Overall, the team tested 80 serial plasma samples (one to four per patient) and were able to create panels incorporating between six and 115 mutations depending on the patient.
Before treatment, the team detected ctDNA in 32 of 32 patients at tumor fractions of 0.002 to 1.06 percent. Plasma samples after completion of neoadjuvant treatment were available for 22 patients, of which 17 tested positive for residual ctDNA, including 12 of 13 patients with invasive or in situ residual disease and 5 of 9 patients with pathologic complete response.
Patients who did have detectable ctDNA after treatment showed tumor fractions ranging from 0.003 percent to 0.045 percent. The investigators also observed that ctDNA concentrations seen after neoadjuvant therapy were more than fivefold lower in patients who achieved pathologic complete response compared to those who didn’t.
In one patient with residual disease, TARDIS failed to detect ctDNA in the last blood sample taken after completion of therapy, which the authors wrote was likely due to a combination of limited plasma DNA available for analysis and a limited number of targets analyzed — just 11.
According to the authors, the results demonstrate a potential application of their described technology: to help avoid overtreatment of early-stage breast cancer patients by better confirming the efficacy of neoadjuvant therapy.
The issue of overtreatment is one that the authors argued is likely to become more relevant as "newer blood- and imaging-based early detection approaches gain credence."
"Our results suggest that blood-based residual disease testing during treatment can further help individualize the choice and extent of each treatment modality," the investigators wrote.
According to Murtaza, he and his colleagues are already working on a clinical validation that will hopefully recapitulate the results they saw in their study this week, as well as "identify and establish quantitative thresholds for minimal residual disease and response."
Having these types of defined cutoffs will be crucial for establishing how this type of monitoring could, or should, be used clinically. And that important clinical utility work, Murtaza said, will be a next step for his team after it completes its clinical validation.
"Future studies could evaluate whether the magnitude of early decrease in ctDNA concentration during neoadjuvant treatment is informative of therapeutic benefit, enabling adaptive treatment designs to rapidly identify systemic treatment options that work for individual patients," the study authors suggested, for example.
Such studies could potentially begin after the group completes this clinical validation, which they hope to do by the end of next year, Murtaza said.